Changes in terrestrial conditions

4.4.1

Snow

Snow is an important element of the global climate system and serves as an reflective cover over Arctic land areas and ice surfaces. It has particular importance for the Barents area (see Chapter 2) where the different economic sectors (transportation, infrastructure, tourism/recreation, hydropower production, agriculture) are affected differently by snow. Snow structure (especially internal ice layers) may affect the Arctic ecosystem and reindeer herding through changes in the nature of the habitat and in access to food for grazing animals. Reindeer grazing affects low vegetation and its effect on snow melt, and there is a link to permafrost, heat fluxes, soil moisture, and run-off. Extreme cold outbreaks usually take place in the presence of snow cover, and there have been suggestions of associations with the northern hemisphere winter circulation (Rutgersson et al., 2015). The snow cover reflects much of the incoming solar radiation and so cools the overlying air, but also acts as an insulator by protecting vegetation from frost-damage. Snow is also a heat sink during snow melt, keeping ground temperature near zero, and on the tundra determining whether vegetation is visible. Snow on the ground is also an important reservoir for some pollutants, and is affected by snowfall, temperature and wind. It has socio-economic implications through hazards in terms of avalanches.

4.4.1.1

Snowfall

Key elements determining snowfall and snow accumulation at any place on land are elevation, latitude and proximity to moisture sources. Moisture access is determined by the general atmospheric and oceanic circulation, as well as local factors (e.g. mountains, lakes, and distance from the coast). Substantial Arctic warming has been observed since the mid-20th century (Bindoff et al., 2013), with a temperature increase of 0.5°C per decade and a 2% increase in precipitation per decade over the past 30 years in the Arctic (Karl et al., 2015). Declining sea ice and increased evaporation are contributing to an increase in atmospheric moisture and thereby to increased Arctic precipitation (Bintanja and Selten, 2014). Screen and Simmonds (2012) found a pronounced decline in historic summer snowfall over the Arctic Ocean and Canadian Arctic Archipelago in the ERA-Interim reanalysis dataset, due to an increase in the proportion of precipitation falling as rain

rather than snow, but little change in total precipitation. They connected the loss of snow on ice to a decrease in surface albedo over the Arctic Ocean, which they found to be comparable to the reduction in albedo associated with the decline in sea ice. The decline in summer snowfall may also explain the thinning of sea ice over recent decades and provides support for the existence of an amplifying feedback associated with warming-induced reductions in summer snowfall.

Local climatic conditions affect precipitation phase, leading to different trends for snowfall (and rain) across the Arctic.

Regions with warmer winter climates, such as Scandinavia and the Baltic Sea Basin, have seen declining snowfall trends (Irannezhad et al., 2016), whereas increasing trends have been reported for regions with colder winter climates, such as Canada and Siberia (Kononova, 2012; Vincent et al., 2015).

Räisänen (2016) used data from 12 RCMs (SRES A1B scenario) from the ENSEMBLES project to project changes in snowfall in northern Europe through the 21st century. Results indicate a decrease in total winter snowfall across almost all of northern Europe by 2069–2099. In contrast, snowfall in the middle of winter is projected to increase in the coldest areas: northern Finland, northern Sweden, northern Norway and the Kola Peninsula. But even in these areas, results indicate a general decline in total annual snowfall. This is due to a decline in the number of snowfall days.

By 2100, the number of snowfall days may be 10–20% lower in northern Fennoscandia, and 20–50% lower in coastal areas.

However, there may be a slight increase (0–10%) in average snowfall intensity during snowfall days. Rutgersson et al. (2015) associated higher snowfall in the Arctic with excess moisture due to warm conditions in the preceding summer and autumn.

4.4.1.2

Snow cover extent

Historical and projected changes in Arctic snow-cover extent (SCE) are reported in the SWIPA update (Brown and Schuler et al., 2017). Changes in spatial, temporal and seasonal SCE reflect changes in drivers such as Arctic warming, Arctic moistening and Arctic greening, and interaction between these drivers and feedback mechanisms. Since 1980 there have been widespread decreases in Eurasian SCE, especially over northern Scandinavia. The ACIA assessment reported a 10% decline in SCE over 30 years, with increasingly shorter snow seasons (ACIA, 2004) associated with increased rates of freezing and thawing. The reinforcing snow albedo feedback has played a key role in the poleward retreat of spring and early summer SCE, and the most pronounced decline in SCE has occurred at high latitudes (60°–70°N) where the potential impacts of snow albedo feedback are greatest. The decline in SCE accelerated between 2007 and 2014, especially in Eurasia where the trend over 1971–2014 was amplified compared 1972–2006. This amplification was mainly due to the stronger decline in SCE over 1971–2014 in spring and early summer (Hernández-Henríquez et al., 2015).

4.4.1.3

Snow cover duration

Over the past 30 to 40 years, snow-cover duration (SCD) has declined by 2 to 5 days per decade in the Arctic, including the Baltic area, mostly due to earlier melt onset in spring (Brown

and Schuler, 2017). Across much of the northern hemisphere, the date of melt onset advanced by about 1–2 weeks over the period 1979–2012 (Mioduszewski et al., 2014). The strongest trends in SCD occurred in northern and western Eurasia.

Arctic coastal and island areas experienced a statistically significant decline in SCD over the period 1978–2007 (Callaghan et al., 2011). There was a statistically significant decline in SCD of about 3 days per decade in Fennoscandia over the period 1951–2007. For the period 1978–2007, there was a statistically significant decline in SCD of 7.3 days per decade over the Fennoscandian sector and 6.3 days per decade over the Barents Sea sector. Rasmus et al. (2014a,b) found the snow season had shortened over the past 30 to 50 years at several observation sites in the reindeer management area of Finnish Lapland. Between 1979 and 2007, melt onset near Sodankylä in northern Finland advanced by 3.4 days per decade, and over northern Fennoscandia, SCD was projected to decrease by 10–15 days under 1°C warming, 15–25 days under 2°C warming, 20–35 days under 3°C warming, and 25–45 days under 4°C warming (Lehtonen et al., 2013). In the Atlantic areas of Russia, Bulygina et al. (2011) found the number of days with snow covering more than 50% of the area surrounding a meteorological station decreased by 1.4%

per decade between 1966 and 2010. Callaghan et al. (2011) found SCD over northern Europe and Siberia has decreased since 1980.

The decline in SCD is projected to be greatest over northern Scandinavia. SCD in 2050 is projected to be 30-40% shorter than in 2011 (Brown and Schuler, 2017), and the expected fall in the annual number of snow cover days in northern Fennoscandia is projected to be greater in coastal regions than mountainous areas (Lehtonen et al., 2013). The main reason for the decline in SCD appears to be earlier melt onset in spring and later freeze-up in autumn. When compared to the mean length of the snow season for 1981–2010, the decrease is expected to vary from 10 to 40 days with a rise in temperature of 1–4°C. Annual SCD is projected to decrease by 10–20% over most of the Arctic by 2055 under the RCP8.5 emissions scenario (Brown and Schuler, 2017) but with much larger decreases (>30%) over the European sector and western Alaska. However, the magnitude and temporal evolution of the projected changes in SCD averaged over the Arctic are strongly dependent on the emission scenario used as the basis for the projected changes in climate.

4.4.1.4

Snow depth and snow water equivalence

Although SCD has broadly decreased across the Arctic, snow depth (SD) and snow-water equivalent (SWE) (mean and maximum values) have shown wide regional variations with both increasing and decreasing trends observed. According to Rasmus et al. (2015), there has been a long-term increase in SD and SCD over most of northern Eurasia. Although maximum SD showed little change in northern Sweden over the period 1905–2003, there has been an increase in mean winter SD of about 2 cm (5%) per decade since 1913 and 10% since 1930–1940 (Rasmus et al., 2015). The duration and maximum thickness of the basal ice layer has decreased in the European part of Russia since 1966. SD over Eurasia increased over the period 1966–2010 over Eurasia (Bulygina et al., 2011).

In Finland and Russia, there have been reports of greater SD but shorter SCD, and SCD is more sensitive to climate change.

Recent updates (Brown and Schuler, 2017) show significant trends for the period 1966–2014 in maximum annual SD in two Russian Arctic regions: the Atlantic Arctic (1.4% per decade) and eastern Siberia (2.4% per decade). This tendency contrasts with that observed in regions with warmer winter climates (e.g. Scandinavia and the Baltic Sea Basin), where the sign and magnitude of the trends in SWE and maximum SD can vary significantly with elevation and distance to the coast.

There are three possible explanations for the recent increase in SD across most of northern Eurasia. First, loss of Arctic sea ice at the start of the cold season has enabled additional water vapor influx into the dry Arctic atmosphere, leading to greater snowfall further south. Second, changes in atmospheric conditions through more intensive cyclonic circulation and more frequent storms have contributed to increased snowfall (Callaghan et al., 2011). Finally, increased precipitation (see Section 4.2.2)

Maximum SD is projected to increase over many areas by 2050, however, the snow season is expected to continue shortening due to the earlier onset of spring melt (Brown and Schuler, 2017). As part of the ENSEMBLES project, Räisänen and Eklund (2012) used regional climate model simulations based on the SRES A1B scenario to project future changes in SWE in northern Europe. They found a general decline in snow amount over the 21st century, but high regional and interannual variability. Individual snow-rich winters may still occur in future decades despite a long-term decline in mean SWE. Climate models project greater changes in future SCD than SWE in the Arctic (Brown and Schuler, 2017). Snow cover in the warmer coastal regions of the Arctic (such as in Alaska and Scandinavia) shows the strongest sensitivities to the warming projected. Only northern Siberia and the Canadian Arctic are projected to see an increase in maximum SD. This contrasting pattern of projected change in maximum SWE is a consequence of the impact of non-linear interactions between rising temperature and increasing precipitation, on snowfall, the snow accumulation period and winter melt events (Räisänen, 2008).

4.4.1.5

Snow quality

Changes in winter climate and especially the frequency and intensity of winter warming events (with or without rain) affect snow properties such as albedo, temperature, density, snow grain size distribution, and ice layers. An observational study by Johansson et al. (2011) of 49 years of snow profile stratigraphy data from Abisko (Sweden), showed an increase in very hard snow layers between 1961 and 2009, with harder snow in early winter and more moist snow during spring. Towards the end of the observation period the number of occasions with very hard snow layers in the snow-pack had more than doubled.

Temperature and precipitation both increased over this period, with the increase in air temperature particularly strong at the start and end of the snow season. Warming events followed by low temperatures increase snow-pack density and can generate ice layers in the snow. These ice layers can impede access to forage for caribou, musk-ox and reindeer (Forchhammer et al.,

1993; Hansen et al., 2014; Vikhamar-Schuler et al., 2016) as well as for small rodents living below the snow (Kausrud et al., 2008). Soil temperature and thus permafrost are also affected by rain-on-snow induced changes in snow properties (Westermann et al., 2011). There are indications that ground ice formation has become more common at the lichen layer in Finland (Rasmus et al., 2015). Using climate model results (2081–2090 and RCP8.5), English et al. (2015) estimated that future net downward short-wave radiation at the top of the atmosphere may increase by 8 W/m² over the Arctic basin due to a decline in surface albedo resulting from a decline in snow and ice cover.

Examples of ecological and societal consequences of rain-on-snow events were reported from Svalbard during and after an extreme event in February 2012 (Hansen et al., 2014).

This resulted in a thick ice layer on the ground, increased permafrost temperatures to 5 m depth, and triggered slush avalanches with major impacts on infrastructure (airport closure, restricted traveling in the terrain, closed roads) and wildlife (reindeer starved because they could not access forage). Future warming may bring more frequent rain-on-snow events (Hansen et al., 2014). The processes leading to hard layers or ground-ice layers occur on daily rather than monthly timescales, and whether specific conditions are problematic depends on the general conditions during winter, not just those on particular days.

4.4.2

Permafrost

The changes taking place in permafrost areas under a warming climate are having various impacts. Thawing permafrost has major consequences for buildings, infrastructure and transport networks designed to be supported by frozen ground. For example, roads can be badly damaged when ice within the soil melts and the land subsides. Another effect of thawing permafrost is the release of methane and the reinforcement of the global greenhouse effect. Thawing permafrost can also increase the risk of erosion and landslides if the frozen water in the soil has been acting as a glue. More details about the present state of the permafrost and its effect on hydrology and vegetation can be found in Chapters 2 and 6. According to the recent SWIPA update, the combination of climate-cryospheric-hydrologic change and multiple ecological feedback processes may cause unpredictable reorganization of ecosystem structure and function, and hence trigger ecosystem shifts or give rise to novel ecosystems (Romanovsky et al., 2017). This tendency has already been observed with vegetation shifts and conversions between terrestrial and aquatic ecosystems. For example, thermokarst lakes and wetlands in ice-rich permafrost environments may drain as the permafrost thaws, resulting in their conversion from aquatic to terrestrial ecosystems (Wrona et al., 2016). Projecting the geographic extent and magnitude of such shifts carries great uncertainty, however.

According to Bring et al. (2016) and Wrona et al. (2016), the aquatic and terrestrial landscapes of the Arctic have experienced many changes in successional patterns and the spatial extent of biomes where tundra has become shrubland and forest. These changes have largely been driven by climate change and changes in hydrology, especially in relation to permafrost thaw and related flow pathways. Such changes are important in terms of

key climate-related fluxes of carbon dioxide, methane, energy, and water (Wrona et al., 2016), and there is an emerging need to establish the spatial extent of ecosystem transformations.

Palsas are frost heaves that contain permanently frozen ice lenses. Norway started monitoring palsa peatlands in 2004 (Hofgaard and Myklebost, 2015), in response to concerns about the consequences of reduced palsas on the ecosystem.

The permafrost is thawing because the Arctic is warming, and is expected to continue to thaw under the projected increases in future temperature. Long-term records at a selection of sites providing good spatial coverage across the Barents area show the permafrost has warmed since the 1990s (Romanovsky et al., 2017). The greatest warming has occurred in the cold permafrost of Svalbard and Russia (Figure 4.15). In northern Russia and in the western Siberian Arctic, temperatures at 10 m depth at cold permafrost sites have increased by ~0.4–0.6°C per decade since the late 1980s, with less warming at warmer permafrost sites (Figure 4.15). The European permafrost is thawing and there has been a northward retreat of the southern boundary of near-surface permafrost in European Russia (Rasmus et al., 2015). Records from Abisko in northern Sweden show the period over which the ground remains frozen each year is decreasing, driven by later freeze-up and earlier spring thaw. Mean monthly air temperature is highly correlated with ground temperature to 100 cm depth, and the warming is correlated with soil surface movement due to freezing and thawing. Rasmus et al. (2015) found the southern limit of patchy near-surface permafrost retreated northward by 20–50 km in European Russia between 1974 and 2008.

Figure 4.15 Time series of ground temperature at depths of 10 to 20 m below the surface at selected measurement sites that fall roughly within the continuous to discontinuous and warm permafrost zones in the Barents area, including Scandinavia, Svalbard, and Russia. Data updated from Christiansen et al. (2010) and Romanovsky et al. (2014, 2015).

Juvvasshøe P30 (20.0m)

Tarfalaryggen P20 (20.0m) ZS-124 (10.0m)

Bolvansky #59 (10.0m) Bolvansky #65 (10.0m)

Bolvansky #56 (10.0m)

Urengoy #15-06 (10.0m) Iskoras Is-B-2 (20.0m) KT-16a (15.0m)

Janssonhaugen P10 (20.0m) Urengoy #15-10 (10.0m)

-7 -6 -5 -4 -3 -2 -1 0

1975 1985 1995 2005 2015

Ground temperature, °C

Although there is a general decrease in permafrost temperature with increasing latitude, this relationship varies between regions. Permafrost is warmer in Scandinavia, Svalbard and northwestern Russia than in other Arctic regions, due to the influence of warm ocean currents and prevailing winds on climate, while elevation is a modifying factor in the Nordic countries (Romanovsky et al., 2010; Sato et al., 2014).

Temporal trends in historical permafrost temperature below the depth of seasonal variation (the top layer of soil that thaws during summer and refreezes in autumn, known as the ‘active layer) in Svalbard and Scandinavia were analyzed by Isaksen et al. (2007b). Updated records from the Nordic monitoring sites show ground temperature at 20 m depth to have increased by 0.3–0.7°C per decade since the late 1990s at the colder mountain permafrost sites (Figure 4.15). A significant temperature increase is measurable to at least 80 m depth, reflecting multi-decadal warming of the permafrost surface.

The high rate of warming on Svalbard since 1998 coincided with a period of higher air temperature. In addition, several extreme and long-lasting warm spells were superimposed on a significant warming trend (Isaksen et al., 2007a; Hansen et al., 2014). Less warming has been observed at warm permafrost sites that have been affected by latent heat exchange close to 0°C. Ground temperature observations at some Nordic sites also confirm permafrost degradation over this period: 1999–2009 (Isaksen et al., 2011) and 2002–2012 (Farbrot et al., 2013).

Active layer thickness (ALT) is more sensitive to short-term variations in climate than deeper ground. ALT records thus exhibit greater interannual variability, mainly in response to variations in summer temperature (e.g. Smith et al., 2009).

Most regions where long-term ALT observations are available show an increase over the past five years (Romanovsky et al., 2015). The Russian European North has been characterized by almost monotonic thickening of the active layer over the past 15 years, reaching a maximum in 2012, but decreasing between 2012 and 2014. In the Nordic countries, records (1996–2013) indicate a general increase in ALT since 1999.

Summer 2014 was particularly warm in the Nordic countries and contributed to the deepest active layer measured to date at some sites (Romanovsky et al., 2017).

McGuire et al. (2016) analyzed uncertainties in the permafrost response to climate change within the permafrost region since 1960 for 15 model simulations. Although all models showed a loss in permafrost area (ALT <3 m) from 1960 to 2009 over the study area (Romanovsky et al., 2017), there were large differences in loss rates among the models. Slater and Lawrence (2013) and Koven et al. (2013) analyzed Earth System Model projections of soil temperature from the CMIP5 database to assess the models’

representation of current-climate soil thermal dynamics. Despite large differences in the extent and rate of change in the permafrost, all models agree that the projected warming and increased snow thickness will result in near-surface permafrost degradation over large areas. In the northern hemisphere, the sensitivity of permafrost to climate change is 0.8–2.3 million km² per 1°C of warming. This range in sensitivity results in a wide range of projections for permafrost loss: 15–87% under the RCP4.5 scenario and 30–99% under RCP8.5. Collectively, the CMIP5 models project that permafrost will have largely disappeared from the present-day discontinuous zone by 2100 under RCP4.5.

4.4.3

Land ice

Whereas tundra dominates the northern Siberian mainland, glaciers and ice caps are mainly located on the Arctic

Whereas tundra dominates the northern Siberian mainland, glaciers and ice caps are mainly located on the Arctic